Abstract:

In recent years, with the development of cloud computing technology, data center network has become a hot topic in industry and academia. Relevant research shows that large flow often carries a large amount of data, and the traditional load balancing method based on ECMP is easy to cause “large flow collision” problem. This paper proposes a virtual slice load balancing method based on Fat-Tree, which uses TSO to segment the virtual stream and optimize the path of the sub-stream after fragmentation. Experiments show that the algorithm is superior to the traditional load balancing method in terms of reducing network propagation delay and improving network throughput.

Key words: data center network, load balancing, virtual fragmentation

摘要：

近年来，随着云计算技术的发展，数据中心网络已成为工业界和学术界的热门话题。相关研究表明，大流通常携带大量数据，传统的基于ECMP的负载均衡方法容易造成“大流碰撞”问题。提出一种基于Fat-Tree的虚拟分片负载均衡方法，利用TSO原理将大流进行虚拟分片处理并对分片后的子流进行路径寻优处理。实验结果表明，该算法无论在降低网络传播时延还是在提高网络吞吐量等方面都优于传统负载均衡方法。

关键词: 数据中心网络, 负载均衡, 虚拟分片